基于ZYNQ的实时水果识别系统
Real-Time Fruit Recognition System Based on ZYNQ
DOI: 10.12677/PM.2023.132030, PDF,    科研立项经费支持
作者: 宋宛容, 向 强*：西南民族大学电子信息学院，四川 成都
关键词: 可编程片上系统；千兆以太网；上位机；YOLO-v3；水果识别；System on a Programmable Chip； Gigabit Ethernet； Host Computer； YOLO-v3； Fruit Recognition
摘要: 由于农业与科技快速发展，机器人被更多用于农业生产中，机器视觉是机器人设计中不可或缺的一部分，如何在嵌入式平台上实现实时视频捕捉与视频处理及优化是目前热门研究方向。本系统采用软硬结合方式实现ZYNQ平台上视频采集水果识别系统。硬件平台为XILINX ZYNQ 7020板卡，双目摄像头采集水果图像并通过以太网口将视频数据传输至上位机；上位机使用改进后的MobileV3-YOLOv3算法对图像进行识别。结果显示，硬件平台能够传送实时视频数据至上位机，且经过预处理IP核处理后的图像有明显的优化，改进后的算法AP50达到92%，FPS为34 f/s，模型参数量为235.61M，并回传识别物体位置三维坐标。
Abstract: Due to the rapid development of agriculture and technology, robots are more used in agricultural production. Machine vision is an indispensable part of robot design. How to realize real-time video capture and video processing and optimization on an embedded platform is currently a popular re-search direction. The system adopts the combination of software and hardware to realize the video collection fruit recognition system on the ZYNQ platform. The hardware platform is XILINX ZYNQ 7020 board, the binocular camera collects fruit images and transmits the video data to the host computer through the Ethernet port; the host computer uses the improved MobileV3-YOLOv3 algo-rithm to identify the images. The results show that the hardware platform can transmit real-time video data to the host computer, and the images processed by the pre-processing IP core are obvi-ously optimized. The improved algorithm AP50 reaches 92%, FPS is 34 f/s, the model parameter quantity is 235.61M, and the position of the object is retrieved three-dimensional.
文章引用：宋宛容, 向强. 基于ZYNQ的实时水果识别系统[J]. 理论数学, 2023, 13(2): 251-264. https://doi.org/10.12677/PM.2023.132030

参考文献

[1] 杨雨诺. 基于PYNQ的图像分类识别技术研究与实现[D]: [硕士学位论文]. 南京: 南京邮电大学, 2020.
[2] 刘培. 人员视频检测智能前端系统设计与实现[D]: [硕士学位论文]. 太原: 中北大学, 2022.
[3] 尼留溪. 枣采摘机器人的目标识别技术研究[D]: [硕士学位论文]. 保定: 河北农业大学, 2021.
[4] Girshick, R., Donahue, J., Darrel, T., et al. (2014) Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation. 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, 23-28 June 2014, 580-587. [Google Scholar] [CrossRef] 
[5] Redmon, J., Divvala, S., Girshick, R., et al. (2016) You Only Look Once: Unified, Real-Time Object Detection. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, 27-30 June 2016, 779-788. [Google Scholar] [CrossRef] 
[6] Redmon, J. and Farhadi, A. (2017) YOLO9000: Better, Faster, Stronger. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, 21-26 July 2017, 7263-7271. [Google Scholar] [CrossRef] 
[7] Redmon, J. and Farhadi, A. (2018) YOLOv3: An In-cremental Improvement. arXiv e-prints, 321-342.
[8] 刘海敏. 基于Zynq的MIMO无线传输系统的设计与研究[D]: [硕士学位论文]. 呼和浩特: 内蒙古大学, 2018.
[9] 车启谣, 严运兵. 基于改进YOLOv3的行人检测研究[J]. 智能计算机与应用, 2022, 12(8): 8-13.

为你推荐